1: %Centre for Mathematical Science, City University\\
2: %Northampton Square, London EC1V 0HB, UK\\
3: %E-mail: \email{A.Fring@city.ac.uk}}
4:
5:
6: \documentclass[proceedings]{JHEP3}
7: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
8: \usepackage{amsfonts}
9: \usepackage{amsmath}
10: \usepackage{epsfig,multicol}
11:
12: \setcounter{MaxMatrixCols}{10}
13: %TCIDATA{OutputFilter=LATEX.DLL}
14: %TCIDATA{Version=4.00.0.2312}
15: %TCIDATA{LastRevised=Tuesday, November 08, 2005 15:36:20}
16: %TCIDATA{<META NAME="GraphicsSave" CONTENT="32">}
17: %TCIDATA{Language=American English}
18:
19: \newbox\mybox
20: \newcommand{\ttbs}{\char'134}
21: \newcommand\fverb{\setbox\mybox=\hbox\bgroup\verb}
22: \newcommand\fverbdo{\egroup\medskip\noindent\fbox{\unhbox\mybox}\ }
23: \newcommand\fverbit{\egroup\item[\fbox{\unhbox\mybox}]}
24: \conference{Integrability of non-Hermitian extensions of CMS-models}
25: \abstract{We consider non-Hermitian but PT-symmetric extensions of
26: Calogero models, which have been proposed by Basu-Mallick and Kundu
27: for two types of Lie algebras. We address the question of whether
28: these extensions are meaningful for all remaining Lie algebras (Coxeter groups)
29: and if in addition one may extend the models beyond the rational case to
30: trigonometric, hyperbolic and elliptic models.
31: We find that all these new models remain integrable, albeit for the
32: non-rational potentials one requires additional terms in the extension
33: in order to compensate for the breaking of integrability.}
34:
35: \title{A note on the integrability of non-Hermitian extensions of
36: Calogero-Moser-Sutherland models}
37: \author{Andreas Fring \\
38: %EndAName
39: Centre for Mathematical Science, City University\\
40: Northampton Square, London EC1V 0HB, UK\\
41: E-mail: \email{A.Fring@city.ac.uk}}
42:
43: \input{tcilatex}
44:
45: \begin{document}
46:
47:
48: \section{Introduction}
49:
50: Traditionally one considers quantum mechanical models and quantum field
51: theories associated with Hermitian Hamiltonians, as in general they are
52: guaranteed to have meaningful energy spectra, lead to conservation of
53: probability densities under time evolution etc. Despite this apparent need
54: for the Hamiltonian to be Hermitian, non-Hermitian Hamiltonian systems have
55: been investigated for some time and found to be physical, as for instance in
56: the context of level crossing \cite{Rot1,Rot2} and 1+1 dimensional quantum
57: field theory \cite{Holl,David}. Fairly recent, the observation that the
58: simple one-particle Hamiltonian with potential term $V=x^{2}(ix)^{\nu }$ for
59: $\nu \geq 0$ possesses a real and positive spectrum \cite{Bender:1998ke} has
60: triggered a sequence of investigations \cite%
61: {Bender:2002yp,Bender:2002vv,Bender:2003fi,Bender:2003gu,Bender:2003ve,Bender:2003wy,
62: Bender:2004by,Bender:2004ej,Dorey:2001hi,Mostafazadeh:2001nr,Mostafazadeh:2002id,
63: Mostafazadeh:2002hb,Mostafazadeh:2002wg,Mostafazadeh:2002pd,Mostafazadeh:2003iz,
64: Mostafazadeh:2003gz,Mostafazadeh:2003qb,Mostafazadeh:2004qh,Mostafazadeh:2004tp,
65: Mostafazadeh:2004mx,Znojil:1999qt,Znojil:2000ia,Znojil:2000fr,Levai:2000di,
66: Znojil:2001ij,Bagchi:2001qu}. One of the outcomes of these studies is the conjecture that Hermiticity is
67: only a sufficient but not a necessary condition for the spectrum to be real
68: and positive. Instead, one may simply demand the Hamiltonian to be $\emph{PT}
69: $-invariant in order to ensure the spectrum to be physical. Inevitably,
70: non-Hermitian Hamiltonians will give rise to various other kinds of
71: problems, such as an indefinite metric \cite{Mostafazadeh:2003iz,Znojil}, so
72: that one is forced to give a proper meaning to unitary evolution etc. This
73: particular problem is overcome by utilizing a new type of symmetry, which
74: seems to be always present when the Hamiltonian is $\emph{PT}$-invariant,
75: and use it to define a new inner-product structure which yields positive
76: definite norms of the associated quantum states \cite%
77: {Bender:2002vv,Bender:2004cbr}. Thus these type of non-Hermitian theories
78: can be made consistent and are not in conflict with concepts of standard
79: quantum mechanics, but can be regarded as meaningful extensions of them.
80: Encouraged by these results similar investigations have also been extended
81: to the realm of quantum field theories \cite%
82: {Frieder,Bender:2004sa,Bender:2004ss,Bender:2005zz,Bender:2005hf}.
83:
84: The sole requirement of $\emph{PT}$-invariance allows to include also
85: various types of momentum dependent terms into the potential of the
86: Hamiltonian, which previously when demanding Hermiticity would have been
87: excluded. Such type of models are attractive as they lead to interesting
88: applications in condensed matter physics, because they are usually of
89: anyonic nature and exhibit generalized exclusion statistics of Haldane type.
90: For instance, one has considered extensions of simple harmonic oscillators
91: of the form $V\sim ixp$ \cite{Swanson,Geyer}, the non-linear Schr\"{o}dinger
92: equation perturbed by higher spatial dispersions \cite{snir}, double delta
93: potentials \cite{Pi}, etc. Motivated by this, Basu-Mallick and Kundu \cite%
94: {Basu-Mallick:2000af} have extended the above mentioned investigations from
95: one to many-particle systems and proposed a new type of model which
96: constitutes a non-Hermitian extension of the rational $A_{\ell }$-Calogero
97: models \cite{Cal2}
98: \begin{equation}
99: \mathcal{H}_{BK}=\frac{p^{2}}{2}+\frac{\omega ^{2}}{2}\sum%
100: \limits_{i}q_{i}^{2}+\frac{g^{2}}{2}\sum\limits_{i\neq k}\frac{1}{%
101: (q_{i}-q_{k})^{2}}+i\tilde{g}\sum\limits_{i\neq k}\frac{1}{(q_{i}-q_{k})}%
102: p_{i}\quad g,\tilde{g}\in \mathbb{R},q,p\in \mathbb{R}^{\ell +1}, \label{BK}
103: \end{equation}%
104: where $p_{i}\equiv -i\partial /\partial q_{i}$. Clearly this Hamiltonian is
105: no longer Hermitian, but its extension remains unchanged when transformed
106: under a time-reversal and a subsequent parity transformation
107: \begin{equation}
108: \emph{P}:~p_{j}\mapsto -p_{j},~~q_{j}\mapsto -q_{j}\qquad \emph{T}%
109: :~p_{j}\mapsto -p_{j},~~q_{j}\mapsto q_{j}~~,~~i\mapsto -i,
110: \end{equation}%
111: i.e.~the Hamiltonian $\mathcal{H}_{BK}$ is $\emph{PT}$-invariant.
112: Subsequently, various aspects of the model have been studied \cite%
113: {Basu-Mallick:2001ce,Basu-Mallick:2003pt,Basu-Mallick:2004ye} and
114: intriguingly it was found that in these models the exclusion and exchange
115: parameter differ, unlike in the conventional Calogero models, that is the
116: case $\tilde{g}=0$, where they are identical.
117:
118: With regard to the standard Calogero models, there are four conceivable
119: generalizations for the Hamiltonian $\mathcal{H}_{BK}$. First a fairly
120: trivial one, a formulation independent of the explicit representation for
121: the roots of the $A_{\ell }$-Weyl group, second a generalization of the
122: possible potentials including those which are trigonometric, hyperbolic and
123: elliptic, third a generalization to Lie algebras (or better Coxeter groups)
124: other than $A_{\ell }$ and fourth the possibility to include more coupling
125: constants. A generalization of (\ref{BK}) to Calogero models of $B_{\ell }$%
126: -type has been studied already in \cite{Basu-Mallick:2001ce}. An \ important
127: question to answer is whether these extended models remain integrable in a
128: similar way as their original counterparts or whether the additional term
129: destroys this valuable property. Despite the fact that the issue of
130: Hermiticity is mainly relevant in the quantum theory, we investigate here
131: the classical integrability of these models. Most likely this will also be
132: important in the quantum theory, as it is well known that in these type of
133: models the quantum theories inherit often many properties of their classical
134: counterparts, especially the feature of being integrable or not. The main
135: purpose of this note is to establish which type of extensions of the
136: Calogero-Moser-Sutherland (CMS) models \cite{Cal2,Mo,Suth1,Per} preserve
137: integrability.
138:
139: \section{Integrability of non-Hermitian PT-invariant extensions of CMS-models%
140: }
141:
142: For simplicity we ignore for the time being the confining term in (\ref{BK}%
143: ), that means we set $\omega =0$, and investigate the following
144: generalization of the Basu-Mallick Kundu model with regard to all four of
145: the above mentioned possible generalizations
146: \begin{equation}
147: \mathcal{H}=\mathcal{H}_{\text{Cal}}+\mathcal{H}_{\text{\emph{PT}}}=\frac{%
148: p^{2}}{2}+\frac{1}{2}\sum\limits_{\alpha \in \Delta }g_{\alpha }^{2}V(\alpha
149: \cdot q)+\frac{i}{2}\sum\limits_{\alpha \in \Delta }\tilde{g}_{\alpha
150: }f(\alpha \cdot q)(\alpha \cdot p). \label{HH}
151: \end{equation}%
152: Here $\Delta $ is any root system invariant under Coxeter transformations.
153: We further assume that the potential and the function $f(x)$ in $\mathcal{H}%
154: _{\text{\emph{PT}}}$ are related as $V(x)=f^{2}(x)$. Besides the rational
155: case $f(x)=1/x$ considered previously for the $A_{\ell }$ and $B_{\ell }$%
156: -case \cite{Basu-Mallick:2000af,Basu-Mallick:2001ce}, we also want to
157: consider the remaining possibilities of the CMS-models, the trigonometric
158: case $f(x)=1/\sin x$, the hyperbolic case $f(x)=1/\sinh x$ and in particular
159: the elliptic case $f(x)=1/\func{sn}x$. The Hamiltonian $\mathcal{H}_{\text{%
160: Cal}}$ in (\ref{HH}) is the usual representation independent, meaning the
161: roots, formulation of the CMS models. The equality of the last term in (\ref%
162: {BK}) and the last term in (\ref{HH}) for $\tilde{g}=$ $g_{\alpha }$ is
163: directly seen when the simple roots of $A_{\ell }$ are expressed in their
164: standard $(\ell +1)$-dimensional representation, see e.g. \cite{Hum}, $%
165: \alpha _{i}=\varepsilon _{i}-\varepsilon _{i+1}$ for $1\leq i\leq \ell $,
166: with $\varepsilon _{i}\cdot \varepsilon _{j}=\delta _{ij}$. Having a
167: formulation independent of the representation of the roots, we can next
168: address the question of how many different coupling constants are permitted.
169: A standard argument is to demand the invariance of the potential under the
170: action of the Coxeter group. As the Coxeter transformations preserve the
171: inner product structure, roots of the same length are mapped into each
172: other, such that the roots can be divided into the two subsets of long and
173: short roots, $\Delta =\Delta _{s}\cup \Delta _{l}$, which are left invariant
174: by the Coxeter transformations. This means based on demanding invariance,
175: the extended models possess also two independent coupling constants like
176: their CMS counterparts
177: \begin{equation}
178: g_{\alpha }=\left\{
179: \begin{array}{c}
180: g_{s}\quad \text{for }\alpha \in \Delta _{s} \\
181: g_{l}\quad \text{for }\alpha \in \Delta _{l}%
182: \end{array}%
183: \right. \qquad \text{and\qquad }\tilde{g}_{\alpha }=\left\{
184: \begin{array}{c}
185: \tilde{g}_{s}\quad \text{for }\alpha \in \Delta _{s} \\
186: \tilde{g}_{l}\quad \text{for }\alpha \in \Delta _{l}%
187: \end{array}%
188: \right. .
189: \end{equation}%
190: Demanding integrability often restricts this choice further, e.g. \cite%
191: {Per,FK,FM}.
192:
193: For the rational version of the Calogero models, i.e.~when $f(x)=1/x$, we
194: note next the crucial property
195: \begin{equation}
196: \eta ^{2}=\alpha _{s}^{2}\tilde{g}_{s}^{2}\sum\limits_{\alpha \in \Delta
197: _{s}}V(\alpha \cdot q)+\alpha _{l}^{2}\tilde{g}_{l}^{2}\sum\limits_{\alpha
198: \in \Delta _{l}}V(\alpha \cdot q)\quad \quad \text{with \ }\eta =\frac{1}{2}%
199: \sum\limits_{\alpha \in \Delta }\tilde{g}_{\alpha }f(\alpha \cdot q)\alpha .
200: \label{cr}
201: \end{equation}%
202: Before using (\ref{cr}), let us first consider an argument to establish that
203: it actually holds. The identity implies that when computing $\eta ^{2}$ all
204: terms involving products of the form $f(\alpha \cdot q)f(\beta \cdot
205: q)(\alpha \cdot \beta )$ for which $\alpha \neq \beta $ cancel each other.
206: To see this we gather all terms in triplets involving two arbitrary roots $%
207: \alpha ,\beta $ and a third root which is their sum $\gamma =\alpha +\beta $%
208: . It may happen though that not all three terms of this type appear in the
209: product $\eta \cdot \eta $ due to the fact that either $\alpha \cdot \beta
210: =0 $ or $\beta \cdot \gamma =0$. In that case we can suitable add several of
211: the missing terms in the hope that overall the additional terms sum up to
212: zero. It turns out that one may always group the terms conveniently and
213: cancel them by means of four basic identities. Keeping our discussion
214: representation independent, these identities can be characterized by the
215: value of the inner product of the co-roots $\hat{\alpha}=2\alpha /\alpha
216: ^{2} $ and $\hat{\beta}=2\beta /\beta ^{2}$. For $\alpha ,\beta ,\gamma
217: =\alpha +\beta $ we find the relations
218: \begin{eqnarray}
219: -\frac{\alpha \cdot \alpha }{(\alpha \cdot q)(\beta \cdot q)}+\frac{\alpha
220: \cdot \gamma }{(\alpha \cdot q)(\gamma \cdot q)}+\frac{\beta \cdot \gamma }{%
221: (\beta \cdot q)(\gamma \cdot q)} &=&0\quad \quad \quad \text{for }\hat{\alpha%
222: }\cdot \hat{\beta}=0, \label{b1} \\
223: \frac{\alpha \cdot \beta }{(\alpha \cdot q)(\beta \cdot q)}+\frac{\alpha
224: \cdot \gamma }{(\alpha \cdot q)(\gamma \cdot q)}+\frac{\beta \cdot \gamma }{%
225: (\beta \cdot q)(\gamma \cdot q)} &=&0\quad \quad \quad \text{for }\hat{\alpha%
226: }\cdot \hat{\beta}=1, \label{b2} \\
227: \frac{\alpha \cdot \beta }{(\alpha \cdot q)(\beta \cdot q)}+\frac{\alpha
228: \cdot \gamma }{(\alpha \cdot q)(\gamma \cdot q)}+\frac{\beta \cdot \beta }{%
229: (\beta \cdot q)(\gamma \cdot q)} &=&0\quad \quad \quad \text{for }\hat{\alpha%
230: }\cdot \hat{\beta}=2, \label{b3} \\
231: \frac{\alpha \cdot \beta }{(\alpha \cdot q)(\beta \cdot q)}+\frac{\alpha
232: \cdot \gamma }{(\alpha \cdot q)(\gamma \cdot q)}-\frac{3\beta \cdot \gamma }{%
233: (\beta \cdot q)(\gamma \cdot q)} &=&0\quad \quad \quad \text{for }\hat{\alpha%
234: }\cdot \hat{\beta}=3, \label{b4}
235: \end{eqnarray}%
236: which may be used successively to establish (\ref{cr}). Relation (\ref{b2}),
237: which applies whenever we have three roots of the same length is most
238: obvious to use as it involves the terms appearing in the sum when computing
239: the product $\eta \cdot \eta $ and it is just a matter of grouping the term
240: together. This involves a non-trivial counting as it requires the precise
241: knowledge of which inner product of the two roots are non-vanishing and also
242: the information that after the re-grouping there are no leftovers.
243: Unfortunately we are not aware of a case independent proof for this.
244: However, we systematically verified this for many Coxeter groups, and based
245: on that we assume that (\ref{cr}) holds in general. To sustain this we
246: present here just some selected examples:
247:
248: As a representative for root systems involving only roots of one length,
249: such as all those related to simply laced Lie algebras, we consider the $%
250: A_{3}$-case. The six positive roots in this case are
251: \begin{equation}
252: \Delta _{A_{3}}^{+}=\{\alpha _{1},\alpha _{2},\alpha _{3},\alpha _{4}=\alpha
253: _{1}+\alpha _{2},\alpha _{5}=\alpha _{2}+\alpha _{3},\alpha _{6}=\alpha
254: _{1}+\alpha _{2}+\alpha _{3}\}.
255: \end{equation}
256: We abbreviate now $\hat{f}_{i}:=\alpha _{i}/(\alpha _{i}\cdot q)$, $\alpha
257: ^{2}=$ $\alpha _{i}^{2}$ for $1\leq i\leq 6$ and $g^{2}=g_{s}^{2}=g_{l}^{2}$%
258: . Then using that the only non-vanishing off-diagonal entries in the Cartan
259: matrix $K_{ij}=2\alpha _{i}\cdot \alpha _{j}/\alpha _{j}^{2}$ are $%
260: K_{12}=K_{21}=K_{23}=K_{23}=-1$, we compute
261: \begin{eqnarray}
262: \eta ^{2} &=&g^{2}\sum\limits_{k=1}^{6}\frac{\alpha ^{2}}{(\alpha _{k}\cdot
263: q)^{2}}+\alpha ^{2}g^{2}\left( \hat{f}_{1}\cdot \hat{f}_{2}+\hat{f}_{1}\cdot
264: \hat{f}_{4}+\hat{f}_{2}\cdot \hat{f}_{4}+\hat{f}_{2}\cdot \hat{f}_{3}+\hat{f}%
265: _{2}\cdot \hat{f}_{5}+\hat{f}_{3}\cdot \hat{f}_{5}\right. \notag \\
266: &&\ \ \ \left. +\hat{f}_{1}\cdot \hat{f}_{5}+\hat{f}_{5}\cdot \hat{f}_{6}+%
267: \hat{f}_{1}\cdot \hat{f}_{6}+\hat{f}_{4}\cdot \hat{f}_{3}+\hat{f}_{4}\cdot
268: \hat{f}_{6}+\hat{f}_{3}\cdot \hat{f}_{6}\right) \\
269: &=&g^{2}\sum\limits_{k=1}^{6}\frac{\alpha ^{2}}{(\alpha _{k}\cdot q)^{2}}.
270: \end{eqnarray}
271: We organized the last terms already successively into triplets in such a way
272: that it is easy to see that they all cancel directly by means of (\ref{b2}).
273:
274: The non-simply laced cases are less straightforward. The simplest example
275: involving long and short roots with $\hat{\alpha}\cdot \hat{\beta}=2$ is the
276: $B_{2}$-case. The four positive roots for this are
277: \begin{equation}
278: \Delta _{B_{2}}^{+}=\Delta _{l}^{+}=\{\alpha _{1},\alpha _{3}=\alpha
279: _{1}+2\alpha _{2}\}\cup \Delta _{s}^{+}=\{\alpha _{2},\alpha _{4}=\alpha
280: _{1}+\alpha _{2}\}.
281: \end{equation}
282: The $B_{2}$-Cartan matrix has entries $K_{12}=-2$ and $K_{21}=-1$, from
283: which we compute
284: \begin{eqnarray}
285: \eta ^{2} &=&\sum\limits_{k=1}^{4}\frac{g_{k}^{2}\alpha _{k}^{2}}{(\alpha
286: _{k}\cdot q)^{2}}+\alpha _{l}^{2}g_{l}g_{s}\left( \hat{f}_{1}\cdot \hat{f}%
287: _{2}+\hat{f}_{1}\cdot \hat{f}_{4}+\hat{f}_{2}\cdot \hat{f}_{3}+\hat{f}%
288: _{3}\cdot \hat{f}_{4}\right) \label{1} \\
289: &=&\sum\limits_{k=1}^{4}\frac{g_{k}^{2}\alpha _{k}^{2}}{(\alpha _{k}\cdot
290: q)^{2}}+\alpha _{l}^{2}g_{l}g_{s}\left( -\frac{\alpha _{2}^{2}}{(\alpha
291: _{2}\cdot q)(\alpha _{4}\cdot q)}+\frac{\alpha _{2}^{2}}{(\alpha _{2}\cdot
292: q)(\alpha _{4}\cdot q)}\right) \label{2} \\
293: &=&\sum\limits_{k=1}^{4}\frac{g_{k}^{2}\alpha _{k}^{2}}{(\alpha _{k}\cdot
294: q)^{2}}.
295: \end{eqnarray}
296: Here we used (\ref{b3}) to obtain the penultimate term in (\ref{2}) from the
297: third and fourth last term in (\ref{1}) and identity (\ref{b1}) to obtain
298: the last term in (\ref{2}) from the last two terms in (\ref{1}). Overall the
299: required terms to complete the identities (\ref{b1}), (\ref{b3}) add up to
300: zero.
301:
302: Identity (\ref{b3}) is required for the $G_{2}$-case. The six positive roots
303: are now
304: \begin{eqnarray}
305: \Delta _{G_{2}}^{+} &=&\Delta _{s}^{+}\cup \Delta _{l}^{+} \\
306: &=&\{\alpha _{1},\alpha _{3}=\alpha _{1}+\alpha _{2},\alpha _{4}=2\alpha
307: _{1}+\alpha _{2}\}\cup \{\alpha _{2},\alpha _{5}=3\alpha _{1}+\alpha
308: _{2},\alpha _{6}=3\alpha _{1}+2\alpha _{2}\}.~~~ \notag
309: \end{eqnarray}
310: The $G_{2}$-Cartan matrix has entries $K_{12}=-1$ and $K_{21}=-3$, which
311: yields
312: \begin{eqnarray}
313: \eta ^{2} &=&\sum\limits_{k=1}^{6}\frac{g_{k}^{2}\alpha _{k}^{2}}{(\alpha
314: _{k}\cdot q)^{2}}+\alpha _{s}^{2}g_{s}^{2}\left( \hat{f}_{1}\cdot \hat{f}%
315: _{3}+\hat{f}_{1}\cdot \hat{f}_{4}+\hat{f}_{3}\cdot \hat{f}_{4}\right)
316: +\alpha _{l}^{2}g_{l}^{2}\left( \hat{f}_{2}\cdot \hat{f}_{5}+\hat{f}%
317: _{2}\cdot \hat{f}_{6}+\hat{f}_{5}\cdot \hat{f}_{6}\right) \notag \label{g1}
318: \\
319: &&+\ \alpha _{l}^{2}g_{s}g_{l}\left( \hat{f}_{1}\cdot \hat{f}_{2}+\hat{f}%
320: _{2}\cdot \hat{f}_{3}+\hat{f}_{1}\cdot \hat{f}_{5}+\hat{f}_{4}\cdot \hat{f}%
321: _{5}+\hat{f}_{3}\cdot \hat{f}_{6}+\hat{f}_{4}\cdot \hat{f}_{6}\right)
322: \label{g2} \\
323: &=&\sum\limits_{k=1}^{6}\frac{g_{k}^{2}\alpha _{k}^{2}}{(\alpha _{k}\cdot
324: q)^{2}}-3\alpha _{l}^{2}g_{s}g_{l}\left( \hat{f}_{1}\cdot \hat{f}_{3}+\hat{f}%
325: _{1}\cdot \hat{f}_{4}+\hat{f}_{3}\cdot \hat{f}_{4}\right) \label{g3} \\
326: &=&\sum\limits_{k=1}^{6}\frac{g_{k}^{2}\alpha _{k}^{2}}{(\alpha _{k}\cdot
327: q)^{2}}.
328: \end{eqnarray}
329: Here we employed (\ref{b2}) to cancel the last two triplets in (\ref{g1}).
330: In the step from (\ref{g2}) to (\ref{g3}) we used (\ref{b4}) and then cancel
331: the last three terms in (\ref{g2}) by means of (\ref{b2}).
332:
333: In a similar manner as for the presented examples we may establish (\ref{cr}%
334: ) for the remaining cases. Unfortunately, we are not aware of a case
335: independent argument to prove this in complete generality. Furthermore, we
336: note that (\ref{cr}) does not hold in general for the non-rational
337: potentials.
338:
339: We return now to our main line of argument and employ the identity (\ref{cr}%
340: ) to re-write the Hamiltonian (\ref{HH}) as a conventional Calogero model
341: with shifted momenta
342: \begin{equation}
343: \mathcal{H}=\frac{1}{2}(p+i\eta )^{2}+\frac{1}{2}\sum\limits_{\alpha \in
344: \Delta }\hat{g}_{\alpha }^{2}V(\alpha \cdot q). \label{hhh}
345: \end{equation}%
346: together with some re-defined coupling constants
347: \begin{equation}
348: \hat{g}_{\alpha }^{2}=\left\{
349: \begin{array}{c}
350: g_{s}^{2}+\alpha _{s}^{2}\tilde{g}_{s}^{2}\qquad \text{for }\alpha \in
351: \Delta _{s} \\
352: g_{l}^{2}+\alpha _{l}^{2}\tilde{g}_{l}^{2}\qquad \text{for }\alpha \in
353: \Delta _{l}%
354: \end{array}%
355: \right. .
356: \end{equation}%
357: Next we recall \cite{Lax} that classical integrability may be established by
358: formulating Lax pair operators $L$ and $M$ as functions of the dynamical
359: variables $q_{i}$ and $p_{i}$, which satisfy the Lax equation $\dot{L}=\left[
360: L,M\right] $, upon the validity of the classical equation of motion
361: resulting from the corresponding Hamiltionian. Taking the observation on
362: board that the extended model and the ordinary Calogero model only differ by
363: a specific shift in the momenta and a re-definition of the coupling
364: constants, it is straightforward to see that this also holds for the Lax
365: operators. Thus we take the conventional Lax operators for the CMS models
366: and simply replace $p\rightarrow p+i\eta $. One may then check directly that
367: \begin{equation}
368: L=(p+i\eta )\cdot H+i\sum\limits_{\alpha \in \Delta }\hat{g}_{\alpha
369: }f(\alpha \cdot q)E_{\alpha }\quad \text{and}\quad M=m\cdot
370: H+i\sum\limits_{\alpha \in \Delta }\hat{g}_{\alpha }f^{\prime }(\alpha \cdot
371: q)E_{\alpha } \label{LM}
372: \end{equation}%
373: fulfills the Lax equation with the constraint
374: \begin{equation}
375: \dot{q}_{j}=\frac{\partial \mathcal{H}}{\partial p_{j}}=p_{j}+i\eta
376: _{j}\qquad \text{and\qquad }\dot{p}_{j}=-\frac{\partial \mathcal{H}}{%
377: \partial q_{j}}=-\frac{\partial \mathcal{H}_{\text{Cal}}}{\partial q_{j}}-i%
378: \dot{\eta}_{j},
379: \end{equation}%
380: if $\dot{L}_{\text{Cal}}=\left[ L_{\text{Cal}},M_{\text{Cal}}\right] $. We
381: choose here as convention the Cartan-Weyl basis commutation relations
382: \begin{equation}
383: \left[ H_{i},H_{j}\right] =0,~~~\left[ H_{i},E_{\alpha }\right] =\alpha
384: ^{i}E_{\alpha },~~~\left[ E_{\alpha },E_{-\alpha }\right] =\alpha \cdot H,~~~%
385: \left[ E_{\alpha },E_{\beta }\right] =\varepsilon _{\alpha ,\beta }E_{\alpha
386: +\beta }. \label{comm}
387: \end{equation}%
388: which is compatible with $\func{tr}(H_{i}H_{j})=\delta _{ij}$, $\func{tr}%
389: (E_{\alpha }E_{-\alpha })=1$. The vector $m$ can be specified as a function
390: of the structure constants $\varepsilon _{\alpha ,\beta }$ and the
391: potential. As all additional terms resulting from the shift $i\eta $ cancel
392: in the Lax equation, the requirements imposed by integrability, i.e.~the
393: validity of the Lax equation are exactly the same as in the non-extended
394: models. Note that the Lax equation is solved directly only for the $A_{\ell
395: } $-algebra, but for other algebras we have to follow the reduction
396: procedures as indicated for instance in \cite{OP6,Per,FK,FM}. Having now
397: established that $L$ and $M$ in (\ref{LM}) are meaningful Lax operators for
398: the extended Hamiltonian (\ref{hhh}), we may compute backwards and expand
399: the kinetic term such that we simply obtain
400: \begin{equation}
401: \mathcal{H}=\frac{1}{2}p^{2}+\frac{1}{2}\sum\limits_{\alpha \in \Delta }\hat{%
402: g}_{\alpha }^{2}V(\alpha \cdot q)+i\eta \cdot p-\frac{1}{2}\eta ^{2}.
403: \label{HHH}
404: \end{equation}%
405: Note that we did not make any assumption on the potential, such that (\ref%
406: {HHH}) is a non-Hermitian \emph{integrable} extension for CMS models for all
407: Coxeter groups, including besides the rational also trigonometric,
408: hyperbolic and elliptic potentials. We observe that when one wishes to
409: preserve integrability for the non-rational potentials one can not simply
410: extend the models by the term $i\eta \cdot p$, but one also has to add the
411: momentum independent term $-\eta ^{2}/2$ in order to compensate for the
412: integrability breaking effect of that term.
413:
414: \section{Conclusions}
415:
416: We have demonstrated that the non-Hermitian extensions for the rational
417: Calogero model proposed by Basu-Mallick and Kundu for the $A_{\ell }$ and $%
418: B_{\ell }$ Coxeter groups can be generalized to all remaining groups in such
419: a way that they are classically integrable. The identity (\ref{cr}) is
420: crucial in this context and it would be interesting to have a rigorous
421: generic, i.e. case independent proof for it. The identity for $\eta ^{2}$
422: ensures that the extended Hamiltonian differs from the original Calogero
423: model only by the one term $i\eta \cdot p$. This simplicity can only be
424: maintained when the potential is rational. However, adding one more term as
425: proposed in (\ref{HHH}) one obtains integrable extensions for all
426: CMS-models. One should stress that the above argument does not exclude yet
427: the possibility that (\ref{HHH}) might be integrable even for non-rational
428: potentials when the last term is dropped. Nonetheless, it establishes that
429: when we include this term they are integrable for sure. It would be very
430: interesting to carry out further studies on these new models along the lines
431: previously followed in \cite%
432: {Basu-Mallick:2001ce,Basu-Mallick:2003pt,Basu-Mallick:2004ye} and beyond.
433:
434: Furthermore, it would be interesting to extend the analysis in \cite%
435: {Brihaye:2003dc}, where the question of solvability ($\neq $ integrability)
436: of some of the discussed models has been addressed.
437:
438: \medskip
439:
440: \noindent \textbf{Acknowledgments}. Discussions with Tanaya Bhattacharyya
441: and Hugh Jones are gratefully acknowledged.
442:
443:
444: \begin{thebibliography}{10}
445:
446: \bibitem{Rot1}
447: I.~Rotter,
448: \newblock A continuum shell model for the open quantum mechanical nuclear
449: system,
450: \newblock Rep. Prog. Phys. {\bf 54}, 635--682 (1991).
451:
452: \bibitem{Rot2}
453: I.~Rotter and A.~F. Sadreev,
454: \newblock Avoided level crossing, diabolic points in the complex plane in a
455: double quantum dot,
456: \newblock Phys. Rev. {\bf E71}, 0362271(14) (2005).
457:
458: \bibitem{Holl}
459: T.~Hollowood,
460: \newblock Solitons in affine Toda field theory,
461: \newblock Nucl. Phys. {\bf B384}, 523--540 (1992).
462:
463: \bibitem{David}
464: D.~I. Olive, N.~Turok, and J.~W.~R. Underwood,
465: \newblock Solitons and the energy momentum tensor for affine Toda theory,
466: \newblock Nucl. Phys. {\bf B401}, 663--697 (1993).
467:
468: \bibitem{Bender:1998ke}
469: C.~M. Bender and S.~Boettcher,
470: \newblock Real Spectra in Non-Hermitian Hamiltonians Having PT Symmetry,
471: \newblock Phys. Rev. Lett. {\bf 80}, 5243--5246 (1998).
472:
473: \bibitem{Bender:2002yp}
474: C.~M. Bender, M.~V. Berry, and A.~Mandilara,
475: \newblock Generalized ${\cal P}{\cal T}$ symmetry and real spectra,
476: \newblock J. Phys. {\bf A35}, L467--L471 (2002).
477:
478: \bibitem{Bender:2002vv}
479: C.~M. Bender, D.~C. Brody, and H.~F. Jones,
480: \newblock Complex Extension of Quantum Mechanics,
481: \newblock Phys. Rev. Lett. {\bf 89}, 270401(4) (2002).
482:
483: \bibitem{Bender:2003fi}
484: C.~M. Bender, P.~N. Meisinger, and Q.-H. Wang,
485: \newblock Calculation of the hidden symmetry operator in $\cal{PT}$-symmetric
486: quantum mechanics,
487: \newblock J. Phys. {\bf A36}, 1973--1983 (2003).
488:
489: \bibitem{Bender:2003gu}
490: C.~M. Bender, P.~N. Meisinger, and Q.-H. Wang,
491: \newblock Finite dimensional PT symmetric Hamiltonians,
492: \newblock J. Phys. {\bf A36}, 6791--6797 (2003).
493:
494: \bibitem{Bender:2003ve}
495: C.~M. Bender, D.~C. Brody, and H.~F. Jones,
496: \newblock Must a Hamiltonian be Hermitian?,
497: \newblock Am. J. Phys. {\bf 71}, 1095--1102 (2003).
498:
499: \bibitem{Bender:2003wy}
500: C.~M. Bender, P.~N. Meisinger, and Q.-H. Wang,
501: \newblock All Hermitian Hamiltonians have parity,
502: \newblock J. Phys. {\bf A36}, 1029--1031 (2003).
503:
504: \bibitem{Bender:2004by}
505: C.~M. Bender and H.~F. Jones,
506: \newblock Semiclassical Calculation of the C Operator in PT-Symmetric Quantum
507: Mechanics,
508: \newblock Phys. Lett. {\bf A328}, 102--109 (2004).
509:
510: \bibitem{Bender:2004ej}
511: C.~M. Bender, S.~F. Brandt, J.-H. Chen, and Q.-h. Wang,
512: \newblock The C operator in PT-symmetric quantum field theory transforms as a
513: Lorentz scalar,
514: \newblock Phys. Rev. {\bf D71}, 065010 (2005).
515:
516: \bibitem{Dorey:2001hi}
517: P.~Dorey, C.~Dunning, and R.~Tateo,
518: \newblock Supersymmetry and the spontaneous breakdown of PT symmetry,
519: \newblock J. Phys. {\bf A34}, L391--L400 (2001).
520:
521: \bibitem{Mostafazadeh:2001nr}
522: A.~Mostafazadeh,
523: \newblock Pseudo-Hermiticity versus PT-Symmetry II: A complete characterization
524: of non-Hermitian Hamiltonians with a real spectrum,
525: \newblock J. Math. Phys. {\bf 43}, 2814--2816 (2002).
526:
527: \bibitem{Mostafazadeh:2002id}
528: A.~Mostafazadeh,
529: \newblock Pseudo-Hermiticity versus PT-Symmetry III: Equivalence of
530: pseudo-Hermiticity and the presence of anti-linear symmetries,
531: \newblock J. Math. Phys. {\bf 43}, 3944--3951 (2002).
532:
533: \bibitem{Mostafazadeh:2002hb}
534: A.~Mostafazadeh,
535: \newblock Pseudo-Hermiticity versus PT symmetry. The necessary condition for
536: the reality of the spectrum,
537: \newblock J. Math. Phys. {\bf 43}, 205--214 (2002).
538:
539: \bibitem{Mostafazadeh:2002wg}
540: A.~Mostafazadeh,
541: \newblock Pseudo-Hermiticity for a Class of Nondiagonalizable Hamiltonians,
542: \newblock J. Math. Phys. {\bf 43}, 6343--6352 (2002).
543:
544: \bibitem{Mostafazadeh:2002pd}
545: A.~Mostafazadeh,
546: \newblock Pseudo-Hermiticity and Generalized PT- and CPT-Symmetries,
547: \newblock J. Math. Phys. {\bf 44}, 974--989 (2003).
548:
549: \bibitem{Mostafazadeh:2003iz}
550: A.~Mostafazadeh,
551: \newblock Is Pseudo-Hermitian Quantum Mechanics an Indefinite-Metric Quantum
552: Theory?,
553: \newblock Czech. J. Phys. {\bf 53}, 1079--1084 (2003).
554:
555: \bibitem{Mostafazadeh:2003gz}
556: A.~Mostafazadeh,
557: \newblock Exact PT-Symmetry Is Equivalent to Hermiticity,
558: \newblock J. Phys. {\bf A36}, 7081--7092 (2003).
559:
560: \bibitem{Mostafazadeh:2003qb}
561: A.~Mostafazadeh,
562: \newblock Pseudo-Unitary Operators and Pseudo-Unitary Quantum Dynamics,
563: \newblock J. Math. Phys. {\bf 45}, 932--946 (2004).
564:
565: \bibitem{Mostafazadeh:2004qh}
566: A.~Mostafazadeh,
567: \newblock PT-Symmetric Cubic Anharmonic Oscillator as a Physical Model,
568: \newblock J. Phys. {\bf A38}, 6557--6570 (2005).
569:
570: \bibitem{Mostafazadeh:2004tp}
571: A.~Mostafazadeh,
572: \newblock Pseudo-Hermitian Description of PT-Symmetric Systems Defined on a
573: Complex Contour,
574: \newblock J. Phys. {\bf A38}, 3213--3234 (2005).
575:
576: \bibitem{Mostafazadeh:2004mx}
577: A.~Mostafazadeh and A.~Batal,
578: \newblock Physical Aspects of Pseudo-Hermitian and $PT$-Symmetric Quantum
579: Mechanics,
580: \newblock J. Phys. {\bf A37}, 11645--11680 (2004).
581:
582: \bibitem{Znojil:1999qt}
583: M.~Znojil,
584: \newblock PT -symmetric harmonic oscillators,
585: \newblock Phys. Lett. {\bf A259}, 220--223 (1999).
586:
587: \bibitem{Znojil:2000ia}
588: M.~Znojil,
589: \newblock Shape invariant potentials with PT symmetry,
590: \newblock J. Phys. {\bf A33}, L61--L62 (2000).
591:
592: \bibitem{Znojil:2000fr}
593: M.~Znojil, F.~Cannata, B.~Bagchi, and R.~Roychoudhury,
594: \newblock Supersymmetry without hermiticity,
595: \newblock Phys. Lett. {\bf B483}, 284--289 (2000).
596:
597: \bibitem{Levai:2000di}
598: G.~Levai and M.~Znojil,
599: \newblock Systematic search for PT symmetric potentials with real energy
600: spectra,
601: \newblock J. Phys. {\bf A33}, 7165--7180 (2000).
602:
603: \bibitem{Znojil:2001ij}
604: M.~Znojil and M.~Tater,
605: \newblock Complex Calogero model with real energies,
606: \newblock J. Phys. {\bf A34}, 1793--1803 (2001).
607:
608: \bibitem{Bagchi:2001qu}
609: B.~Bagchi, C.~Quesne, and M.~Znojil,
610: \newblock Generalized Continuity Equation and Modified Normalization in
611: PT-Symmetric Quantum Mechanics,
612: \newblock Mod. Phys. Lett. {\bf A16}, 2047--2057 (2001).
613:
614: \bibitem{Znojil}
615: M.~Znojil,
616: \newblock Conservation of pseudo-norm in ${\cal P}{\cal T}$ symmetric quantum
617: mechanics,
618: \newblock Rendic. Circ. Mat. Palermo {\bf 72}, 211--218 (2004).
619:
620: \bibitem{Bender:2004cbr}
621: C.~M. Bender, J.~Brod, A.~Refig, and M.~Reuter,
622: \newblock The $\cal{C}$-operator in $\cal{PT}$-symmeric quantum theories,
623: \newblock J. Phys. {\bf A37}, 10139--10165 (2004).
624:
625: \bibitem{Frieder}
626: F.~Kleefeld,
627: \newblock Non-Hermitian quantum theory and its holomorphic representation:
628: Introduction and applications,
629: \newblock Los Alamos preprint hep-th/0408097 .
630:
631: \bibitem{Bender:2004sa}
632: C.~M. Bender, D.~C. Brody, and H.~F. Jones,
633: \newblock Extension of PT-symmetric quantum mechanics to quantum field theory
634: with cubic interaction,
635: \newblock Phys. Rev. {\bf D70}, 025001(19) (2004).
636:
637: \bibitem{Bender:2004ss}
638: C.~M. Bender,
639: \newblock PT symmetry in quantum field theory,
640: \newblock Czech. J. Phys. {\bf 54}, 13--28 (2004).
641:
642: \bibitem{Bender:2005zz}
643: C.~M. Bender, I.~Cavero-Pelaez, K.~A. Milton, and K.~V. Shajesh,
644: \newblock PT-symmetric quantum electrodynamics,
645: \newblock Phys. Lett. {\bf B613}, 97--104 (2005).
646:
647: \bibitem{Bender:2005hf}
648: C.~M. Bender, H.~F. Jones, and R.~J. Rivers,
649: \newblock Dual $\cal{PT}$-symmetric quantum field theories,
650: \newblock Phys. Lett. {\bf B625}, 333--340 (2005).
651:
652: \bibitem{Swanson}
653: M.~S. Swanson,
654: \newblock Transition elements for a non-Hermitian quadratic Hamiltonian,
655: \newblock J. Math. Phys. {\bf 45}, 585--601 (2004).
656:
657: \bibitem{Geyer}
658: H.~B. Geyer, F.~G. Scholtz, and I.~Snyman,
659: \newblock Quasi-hermiticity and the role of a metric in some boson
660: Hamiltonians,
661: \newblock Czech. J. Phys. {\bf 54}, 1069--1073 (2004).
662:
663: \bibitem{snir}
664: A.~G. Shnirman, B.~A. Malomed, and E.~Ben-Jacob,
665: \newblock Nonperturbative studies of a quantum higher-order nonlinear
666: Schrodinger model using the Bethe ansatz,
667: \newblock Phys. Rev. {\bf A50}, 3453--3463 (1994).
668:
669: \bibitem{Pi}
670: U.~Aglietti, L.~Griguolo, R.~Jackiw, S.-Y. Pi, and D.~Seminara,
671: \newblock Anyons and Chiral Solitons on a Line,
672: \newblock Phys. Rev. Lett. {\bf 77}, 4406--4409 (1996).
673:
674: \bibitem{Basu-Mallick:2000af}
675: B.~Basu-Mallick and A.~Kundu,
676: \newblock Exact solution of Calogero model with competing long-range
677: interactions,
678: \newblock Phys. Rev. {\bf B62}, 9927--9930 (2000).
679:
680: \bibitem{Cal2}
681: F.~Calogero,
682: \newblock Solution of a three-body problem in one-dimension,
683: \newblock J. Math. Phys. {\bf 10}, 2191--2196 (1969).
684:
685: \bibitem{Basu-Mallick:2001ce}
686: B.~Basu-Mallick and B.~P. Mandal,
687: \newblock On an exactly solvable $B_N$ type Calogero model with nonhermitian PT
688: invariant interaction,
689: \newblock Phys. Lett. {\bf A284}, 231--237 (2001).
690:
691: \bibitem{Basu-Mallick:2003pt}
692: B.~Basu-Mallick, T.~Bhattacharyya, A.~Kundu, and B.~P. Mandal,
693: \newblock Bound and scattering states of extended Calogero model with an
694: additional PT invariant interaction,
695: \newblock Czech. J. Phys. {\bf 54}, 5--12 (2004).
696:
697: \bibitem{Basu-Mallick:2004ye}
698: B.~Basu-Mallick, T.~Bhattacharyya, and B.~P. Mandal,
699: \newblock Phase shift analysis of PT-symmetric nonhermitian extension of
700: $A_{N-1}$ Calogero model without confining interaction,
701: \newblock Mod. Phys. Lett. {\bf A20}, 543--552 (2005).
702:
703: \bibitem{Mo}
704: J.~Moser,
705: \newblock Three integrable Hamiltonian systems connected with isospectral
706: deformations,
707: \newblock Adv. Math. {\bf 16}, 197--220 (1975).
708:
709: \bibitem{Suth1}
710: B.~Sutherland,
711: \newblock Quantum many body problem in one-dimension: Ground state,
712: \newblock J. Math. Phys. {\bf 12}, 246--250 (1971).
713:
714: \bibitem{Per}
715: A.~M. Perelomov,
716: \newblock Integrable Systems of classical Mechanics and Lie Algebras, Vol I,
717: \newblock Birkh{\"a}user Verlag, Berlin (1990).
718:
719: \bibitem{Hum}
720: J.~E. Humphreys,
721: \newblock Introduction to Lie Algebras and Representation Theory,
722: \newblock Springer, Berlin (1972).
723:
724: \bibitem{FK}
725: A.~Fring and C.~Korff,
726: \newblock Non-crystallographic reduction of generalized Calogero-Moser models,
727: \newblock CMS 07/05, hep-th/0509152 .
728:
729: \bibitem{FM}
730: A.~Fring and N.~Manojlovic,
731: \newblock $G_{2}$-Calogero-Moser Lax operators from reduction,
732: \newblock CMS 08/05, hep-th/0510012 .
733:
734: \bibitem{Lax}
735: P.~Lax,
736: \newblock Integrals of nonlinear equations and solitary waves,
737: \newblock Commun. Pure Appl. Math. {\bf 21}, 467--490 (1968).
738:
739: \bibitem{OP6}
740: M.~A. Olshanetsky and A.~M. Perelomov,
741: \newblock Completely integrable Hamiltonian systems connected with semisimple
742: Lie algebras,
743: \newblock Invent. Math. {\bf 37}, 93--108 (1976).
744:
745: \bibitem{Brihaye:2003dc}
746: Y.~Brihaye and A.~Nininahazwe,
747: \newblock On PT-symmetric extensions of the Calogero model,
748: \newblock Int. J. Mod. Phys. {\bf A19}, 4391--4400 (2004).
749:
750: \end{thebibliography}
751:
752:
753: %\bibliographystyle{phreport}
754: %\bibliography{Ref}
755:
756: \end{document}
757: